Thanks.  I have been looking at the _cluster.sl_status table, and it's 
been really useful.  The only problem I've got now is I would like to be 
able to do some calculations with the st_lag_time field in 
_my_cluster.sl_status table.  As it is stored as an Interval, I would 
really like it in just one format (e.g. seconds) so I could use it as an 
integer.  At the moment it looks something like this...

ipt=# select st_lag_time from _my_cluster.sl_status;
   st_lag_time
-----------------
 00:00:02.114107
(1 row)

I have looked into converting Interval Types, but have been told that 
this conversion will no longer be supported by Postgres.

Any ideas?

Regards,

Aaron


Christopher Browne wrote:
> Aaron Randall <aaron.randall at visionoss.com> writes:
>   
>> Hi Everyone :)
>>
>> I have a question similar to one that was raised a little while ago titled - "How do I know when a subscriber has caught up?".
>>
>> Is there a flag somewhere that confirms that the master and slave
>> are connected (regardless of whether they are currently transmitting
>> data) to show that Slony is running successfully on both.  At the
>> moment, the only was I can check is by tailing the Slony logs, but
>> ideally I would like to use some kind of database field that could
>> confirm/deny the master and slave(s) are connected successfully.
>>     
>
> Well, you can get some implicit "all is running well" if you run
>   select * from _cluster.sl_status;
>
> If it shows all the nodes you expect, and shows that they are only an
> event or two behind, then replication should be functioning well.
>
> That's about the best metric I can think of without looking at the
> logs.
>
> You could also run the script test_slony_state.pl (or the DBI
> version); that checks a number of things about how things are running.
>